Dissecting the Molecular Response of Human Escs to Iron-Mediated Oxidative Stress by Genetic Silencing of FTH1 Gene


 Background: Embryonic stem cells (ESCs) are pluripotent cells with indefinite self-renewal ability and differentiation properties. As such, to function properly and maintain genomic stability, ESCs need to be endowed with an efficient repair system as well as effective redox homeostasis. In this study, we investigated and characterized different aspects involved in ESCs response to iron accumulation following stable knockdown of ferritin heavy chain (FTH1) gene, encoding for a major iron storage protein with ferroxidase activity. Methods: stable FTH1 knockdown of H9-hES cell line was achieved with the use of shRNA lentiviral particles. Upon FTH1 silencing, we speculated whether hESCs still retained their pluripotency capability were first monitored for their capability the pluripotent status,FTH1 stable knock-down ESCs were obtained using lentiviral vector plasmids. The effect of FTH1 silencing on hESCs pluripotency was evaluated through alkaline phosphatase (AP) staining, immunofluorescence and embryoid bodies (EBs) formation assay. Western blotting and qRT-PCR analysis were performed to assess the involvement of nuclear factor (erythroid-derived-2)-like 2 (Nrf2) and pentose phosphate pathway (PPP) in the antioxidant response. ROS levels and mitochondrial functionality were explored by flow cytometry. Seahorse Analyzer was used to evaluate metabolic and bioenergetic profiles. Results: Our findings clearly show that FTH1 silencing in hESCs does not correlate with increased ROS production nor with redox status strengthening the concept that hESCs are extremely resistant and, to certain extent, even refractory to the pattern of results produced in other cell lines. Collectively, our results demonstrate that FTH1 silencing is accompanied by a significant activation of the nuclear factor (erythroid-derived-2)-like 2 (Nrf2) signaling pathway and pentose phosphate pathway (PPP) which crosstalk in driving hESCs antioxidant cascade events able to antagonize the effects of FTH1 silencing.Conclusion: to our knowledge, this is the first evidence of a crosstalk between FTH1 silencing and Nrf2 pathway activation in hESCs, casting a new light on how human ESCs perform under oxidative stress conditions. Our findings go beyond previous reports, showing how the Nrf2 pathway, in combination with PPP activation, regulates the molecular signature underlying ESCs defence mechanisms against oxidative stress mediated by FTH1 downregulation.


Introduction
An impaired cellular balance between reactive oxygen species (ROS) production and clearance is responsible for establishing an oxidative stress status leading to lethal consequences (18) due to the toxic effects responsible for damage in many cellular components, including DNA. Depending on time and concentration, ROS elicit a wide spectrum of responses; indeed, slight variation in ROS concentrations may have profound and opposite effects, such as cell survival and apoptosis (49). High levels of ROS, predominantly produced by the mitochondrial respiration, are classically linked to cellular damage, oxidative stress, DNA damage, and apoptosis, highlighting the importance of proper mitochondrial biogenesis and reduced oxygen consumption (17,62,85). The antioxidant system includes a wide range of enzymes such as those encoded by the superoxide dismutase family genes (SODs), glutathione peroxidase family (GPXs), and catalase (CAT). In addition, many other non-enzymatic molecules and compounds such as ascorbic acid, directly or indirectly, participate in antioxidant events (11). The redox status is greatly in uenced by compounds primarily involved in the metabolism of speci c substances, such as iron (20). Iron de ciency triggered by Deferoxamine (DFO) treatment was shown to be involved in loss of pluripotency both in ESCs and induced pluripotent stem cells (iPSCs) (25). On the other hand, increased intracellular iron levels also negatively affect the status of pluripotent stem cells (PSCs), leading to excess ROS production, DNA damage, and reduced proliferation rate (24). The role of ROS in sophisticated and nely regulated mechanisms such as pluripotency and differentiation has been described in many studies, but the real effects of ROS accumulation in ESCs and the underlying molecular mechanisms still remain imprecise; moreover, even less speculated is the role of unbalanced iron metabolism in hESCs, suggesting that much still remains to be investigated in this direction. PSCs mainly rely on glycolysis (42) while oxidative metabolism (mitochondrial respiration and fatty acid metabolism) take place upon PSCs differentiation (17,46,86). Reprogramming of somatic cells to generate iPSCs requires, among others, a metabolic switch from oxidative phosphorylation (OXPHOS) to glycolysis, strengthening the high dependency of PSCs on glycolytic metabolism (62). The prevailing aerobic over oxidative mitochondrial metabolism in PSCs is reminiscent of cancer metabolism (7, 28, 45, 62). As cancer cells, in fact, PSCs divide rapidly and, in order to sustain the high proliferation rate, these cells are required to produce ATP from glycolysis even in the presence of normal oxygen levels. Glycolytic metabolism also generates nicotinamide adenine dinucleotide phosphate, NADPH that, in addition to being involved in sustaining the high proliferation rate, protects against oxidative stress (48).
Additionally, as cancer cells, PSCs are characterized by an increased expression of hypoxia-inducible factors (HIFs) and oxygen-sensitive transcription factors. Nevertheless, oxidative mitochondrial metabolism was shown to be active in hESCs in particular conditions, such as high lipids, and metabolic ux analysis (MFA) demonstrating that PSCs can adapt their metabolism to various nutrient conditions while maintaining full self-renewal properties (84).
To push the PSCs technology towards clinical applications, a comprehensive understanding of stemness and its potentiality, as well as stem cells metabolism and its modulation are fundamental steps to be considered to generate PSC derivatives that faithfully recapitulate all aspects of somatic cells and tissues. While ESCs general metabolic aspects have been widely investigated (43,77,84), very little is currently known about the role of iron metabolism and the consequences of iron accumulation in hESCs.
Based on these considerations, the present work speculates the molecular events triggered by silencing of the FTH1 gene, encoding a major antioxidant protein with ferroxidase activity. We therefore analyzed key processes known to be associated with intracellular iron accumulation such as DNA damage, apoptosis, oxidative stress, antioxidant defence system, and metabolic changes. Our ndings provide clear evidence of overactivation of the nuclear factor (erythroid-derived-2)-like 2 (Nrf2) signaling pathway as the major player and safeguard of hESCs against iron-mediated oxidative stress in response to downregulation of ferritin expression. From a metabolic point of view, we identi ed the pentose phosphate pathway (PPP) activation as the primary metabolic response to maintain the redox status in hESCs.

Lentiviral production and generation of stable knock-down ESCs
Lentiviral particles were produced in 293T cells following standard procedure. Brie y, subcon uent 293T cells were transfected with shFTH1 or shSCR control lentiviral vector plasmids (Sigma-Aldrich, St. Louis, MO) along with pCMV-DR9-9 and pCMV-VSVG plasmids (Sigma-Aldrich). After 48 hours, supernatants were harvested, ltered through a 0.45 µM lter, and used for ESCs infection, previously seeded in a 6-well plate at a cell density of 2x10 5 cells/well. Infected cells were incubated with shFTH1 and shSCR control lentiviral particles [Multiplicity of Infection (MOI) = 10] in the presence of 8 µg/ml of Polybrene (Sigma-Aldrich) using Lipofectamine 3000 Reagent (Thermo Fisher Scienti c, Waltham, MA). Two days after transduction, transduced cells were selected with puromycin (1µg/ml) for 10 days. Knock-down e ciency was assessed via both western blot and qRT-PCR analysis (Fig. S1).

Measurement of intracellular ROS and mitochondrial superoxide species
Total intracytoplasmic ROS as well as speci c mitochondrial superoxide species were assessed by ow cytometry using the uorescent probes 2′-7′-Dichlorodihydro uorescein diacetate (CM-H2DCFDA; Thermo Fisher Scienti c) and MitoSOX™ Red Mitochondrial Superoxide Indicator, respectively. For the quanti cation of mitochondrial superoxide levels, cells were incubated with 2mM MitoSOX™ Red Mitochondrial Superoxide Indicator (Thermo Fisher Scienti c) for 30 minutes at 37°C. Intracellular ROS levels were instead assayed by staining the cells with 1,5 mM CM-H2DCFDA (Thermo Fisher Scienti c) for 30 minutes at 37°C. Fluorescence intensity was measured by ow cytometry using a FACS BD LSRFortessaTM X-20 cyto uorimeter (BD Biosciences, Bedford, MA). The results were analyzed with FlowJo software (Tree Star, Inc.). Three independent experiments were conducted.

Analysis of mitochondrial membrane potential (ΔΨ)
Mitochondrial membrane potential (ΔΨ) was measured by using 0,1 ΔΨ uorescent indicator TMRM (tetramethylrhodamine methyl ester) (Thermo Fisher Scienti c). Fluorescence intensity was measured after 30 min of incubation at 37°C by ow cytometry using a FACS BD LSRFortessaTM X-20 cyto uorimeter (BD Biosciences). The results were analyzed with FlowJo software (Tree Star, Inc.). Three independent experiments were conducted.
1.5 x 10 6 cells were resuspended in Brew medium containing 10 µM Y27632 ROCK inhibitor (Selleckchem, S1049), and cultured onto ultra-low attachment plates (TPP, 93060). Cells were maintained in these conditions for 3 days. After 7 days, oating EBs were plated on laminin-coated dishes Rate Assay Kit. Simultaneous quanti cation of ATP fractions produced by mitochondrial oxidative phosphorylation (OXPHOS) and glycolysis were detected. For bioenergetics quanti cation, cells were dissociated into single cells with Accutase and seeded at a cell density of 2 x 10 4 /well in Brew medium supplemented with 10 mM Y27632 ROCK inhibitor. The day after, the growth medium was exchanged with unbuffered media and incubated for 30 minutes at 37°C allowing temperature and pH to reach the equilibrium before starting the baseline measurements. Once the basal oxygen consumption rate (OCR) and extracellular acidi cation rate (ECAR) were obtained, cells were metabolically perturbed by the additions of three different compounds: Oligomycin (1.5 µM), Rotenone + Antimycin A (0.5 µM). Finally, the template was loaded into the Seahorse Analyzer for ATP measurement. The same procedure for cell preparation was used for the Glycolytic rate assay and measurement performed according to the manufacturer's instruction. Rotenone/antimycin A were used to block mitochondrial activity while 2-DG to inhibit glycolysis. Results were uploaded to Agilent Seahorse Analytics to calculate parameters relative to ATP production and XF Glycolytic Rate assays. GraphPad Prism software was used to create graphs.

Statistics
All statistical data were generated from at least three independent biological replicates and are represented as mean ± standard error of the mean (SEM). Data were analyzed with PRISM version 7.0 (GraphPad Software Inc., San Diego, CA, USA). The signi cance of differences (p-value) was analyzed using Student's t-test and presented with the following levels of signi cance: *P < 0.05, **P < 0.01, and ***P < 0.001.

FTH1-silencing mediates impairment of ESCs pluripotency.
Iron homeostasis is involved in many biological processes including the maintenance of pluripotency in human PSCs. For instance, results from previous research found that depletion of intracellular iron associates with impairment of pluripotency and self-renewal due to signi cant reduction of NANOG expression (25). Cells maintain a balanced iron pool and in this regulation ferritin, as iron storage protein, plays a crucial role. Therefore, to speculate on the role of ferritin in the context of hESCs, we rst generated FTH1-silenced cells via lentiviral transfection with short hairpin RNA (shRNA); subsequent analysis at both mRNA and protein levels con rmed that FTH1 was successfully knocked-down in transfected hESC ( Supplementary Fig. S1B-S1E). Morphologically, colonies from FTH1-KD hESCs were more attened than those of shSCR control ( Supplementary Fig. S1A), and alkaline phosphatase (AP) activity was reduced in FTH1-silenced cells (Fig. 1A). FTH1-KD hESCs colonies were positive for pluripotency markers Oct4, TRA-1-60, and Sox2, while the expression of Nanog was slightly increased ( Fig. 1B and Supplementary Fig. S2A). This last nding was further con rmed by immunoblot analysis demonstrating that Nanog expression is indeed higher in FTH1-silenced hESCs. (Fig. 1C). At the transcriptional level, OCT4, SOX2, and NANOG resulted evenly increased in FTH1-KD hESCs (Fig. 1D). We further asked whether FTH1 knockdown could, either directly or indirectly, promote spontaneous differentiation of hESCs. Analysis of the transcript levels of AFP, Nkx2.5, and PAX6, speci c of endoderm, mesoderm, and ectoderm germ layers respectively, revealed an enhanced expression of these genes in FTH1-KD hESCs (Fig. 1E, left panel). Next, we induced both SCR control and FTH1-KD hESCs to differentiate using the Embryoid bodies (EBs) formation method (Supplementary S2B). On day 28 of differentiation, EBs were harvested and analyzed for endodermal-(GATA4, FOXA2), mesodermal-(HAND1, CD31), and ectodermal-(NESTIN, PAX6) associated markers. While endodermal and mesodermal transcript levels resulted upregulated in FTH1-KD-derived EBS compared to control EBs, the ectodermal genes NESTIN and PAX6 were signi cantly downregulated in EBs derived from silenced cells (Fig. 1E, right panel), suggesting that modulation of FTH1 by repression negatively affects neuroectodermal gene expression. Immuno uorescence analysis for Sox17 (endodermal marker), Brachyury (mesodermal marker), and Nestin further con rmed that the percentage of Nestin + cells is lower in FTH1-silenced hESCs compared to SCR control (Fig. 1F). A similar pattern of results was observed in a previous study, in which overexpression of NANOG was associated with neuroectodermal differentiation impairment of hESCs (80).

ROS levels are reduced upon FTH1-gene silencing in hESCs
The role of ferritin in preventing iron-mediated oxidative stress was observed in many studies reporting a direct correlation between modulation of FTH1 expression and dysregulation of iron homeostasis (75). Speci cally, a reduction in FTH1 expression leads to excess labile iron, in turn promoting the formation of oxygen-derived free radicals (39,67). To speculate on the effects of FTH1 silencing in hESCs, we measured total and mitochondrial ROS levels together with an analysis of the mitochondrial membrane potential (ΔΨM) on three biological replicates of SCR and FTH1-KD hESCs, using DCF, MitoSox, and TMRM dyes, respectively (Supplementary S3A and S3B). Intriguingly, our data revealed a reduction in total ROS levels in FTH1-silenced hESCs compared to SCR controls (FTH1-KD: CM-H2DCFDA MFI: 29785,7 vs SCR: CM-H2DCFDA MFI: 47341,7) ( Fig. 2A). Likewise, MitoSox assay highlighted a signi cant reduction in mitochondrial superoxide in FTH1-de cient cells (FTH1-KD: MitoSOX Red MFI: 251,8 vs SCR: MitoSOX Red MFI: 1002,3) (Fig. 2B). A similar, asymmetric behavior was observed by ow cytometry analysis, which revealed a decrease in TMRM uorescence, indicative of mitochondrial membrane depolarization, in FTH1-KD hESCs (MFI: 7722) compared to SCR control (MFI: 11086,7) (Fig. 2C). Together, these results led us to conclude that hESCs had to operate and overactivate well established antioxidant systems in order to protect themselves from shFTH1-mediated iron toxicity.

Nrf2 signaling pathway drives the antioxidant response in hESCs
The nuclear factor (erythroid 2-related) factor 2 (Nrf2) is a transcription factor ubiquitously expressed in most eukaryotic cells. Nrf2-Keap1 (Kelch-like ECH-associated protein 1) signaling pathway was shown to play a central role in protecting cells against oxidative stress (41,50). Under basal conditions, Nrf2 is bound to its negative regulator Keap1 which directs it to proteasomal degradation. In the presence of intracellular ROS, Keap1 is oxidized and releases Nrf2 which is free to translocate into the nucleus, where it binds to antioxidant response elements (AREs) and induces the expression of antioxidant target genes, such as HMOX1, NQO1, GST, GPX (36). Besides its role as a ne regulator of redox and metabolic homeostasis, Nrf2 also acts as a pluripotency master gene: Nrf2 activation in hESCs was shown to enhance NANOG transcriptional activity by delaying Nanog protein degradation through POMP-mediated proteasome ubiquitination (27,33). We found that FTH1-KD hESCs express higher levels of Nrf2 transcript (Fig. 3A) and protein ( Fig. 3B and 3C; Supplementary Fig. S4A), assessed via qRT-PCR, immunoblot, and immuno uorescence ( Supplementary Fig. S4F), respectively. Similarly, we could also observe an increased expression of p62/Sqstm1 (Fig. 3D and Supplementary Fig. S4B) which, in the noncanonical Keap1-Nrf2 pathway, is known to bind Keap1, allowing Nrf2 to migrate into the nucleus (37). Moreover, the p62 protein yields a feedback loop that ampli es the Nrf2 system (32). In addition to increased expression of Nrf2 in FTH1-silenced hESCs, we observed the overexpression of some important antioxidant enzymes, such as glutathione peroxidases Gpx2, Gpx3 (Fig. 3E), Gpx4 ( Fig. 3F and Supplementary Fig. S4C), and superoxide dismutases (Sod) 1 and 2 (Fig. 3G). Also, more Nrf2-regulated genes (HMOX1, NQO1, HIF1 , HIF2 ) (Fig. 3H) and proteins (Hif1 , Ftl) ( Fig. 3I and Fig. 3L, respectively and Supplementary Fig. S4D and S4E) were found highly expressed in FTH1-KD hESCs. Altogether, these ndings suggest a crosstalk between FTH1 silencing and overactivation of the Nrf2 signaling pathway and its cognate effectors in hESCs.

FTH1 knock-down triggers Apoptosis and DNA damage
Labile iron accumulation is a well-known cell damage effector and pro-apoptotic factor (13,35,74); intracellular iron overload by ferric ammonium citrate (FAC) treatment leads to ferroptosis, an ironregulated cell death (15). To check the effects of FTH1-silencing on apoptosis, expression analysis of intrinsic pro-apoptotic genes such as BAX and BIM, as well as CASP9 and CASP3 was performed, revealing a signi cant upregulation of their expression in FTH1-KD with respect to SCR control hESCs ( Fig. 4A and 4B). Similarly, immunoblot analysis of cleaved Casp3 protein con rmed its overexpression in silenced hESCs ( Fig. 4C and Supplementary Fig. S5A). It is by now generally accepted that Akt and Erk1/2 pathways are both linked to apoptosis, but with opposite effects: Erk promotes apoptosis, both intrinsically and extrinsically and its activity can be driven by the presence of ROS (90). By contrast, Akt is linked to cell survival, therefore its activity mediates suppression of apoptosis (38). Based on these observations, we speculated whether Erk and Akt could be involved in our FTH1-KD hESCs system; therefore, we measured their expression levels by immunoblot analysis and found that the active, phosphorylated Akt protein (pAkt) expression was reduced, while the expression of pErk1/2 resulted signi cantly increased in FTH1-silenced hESCs ( Fig. 4D and 4E, respectively; Supplementary Fig. S5B and S5C). Collectively, these ndings clearly suggest that an active cell death program occurs in hESCs when ferritin is stably downregulated. In addition, based on the evidence that high ROS levels and FTH1 modulation also associate with DNA damage (8, 65, 73, 83), we measured the expression of some DNA damage-associated genes. In line with previous studies, here we show that BRCA1, DMC1, PCNA, and POLQ are indeed signi cantly upregulated in FTH1-KD hESCs (Fig. 4F). Similarly, we could also observe an increased expression of Parp1 protein, a substrate of activated caspase 3, in FTH1-silenced hESCs ( Fig. 4G and Supplementary Fig. S5D). Overall, these results provide strong evidence that FTH1 silencing induces apoptosis in hESCs and triggers the activation of DNA-damage response programs.

Effects of FTH1 repression on cellular metabolism
Under oxidative stress conditions, a metabolic shift from OXPHOS to glycolysis was reported to occur (87), suggesting a clear and intimate correlation between oxidative stress and metabolic changes. It is also well known that hPSCs rely on glycolysis while OXPHOS takes place as soon as these cells enter differentiation (77,79). By using the Agilent Seahorse analyzer, we rst quanti ed the fraction of ATP generated by glycolysis and by OXPHOS. The difference between SCR control and FTH1-silenced hESCs in glycolysis vs. OXPHOS ATP fraction was negligible. However, we noticed an overall reduction of total ATP production from both sources ( Fig. 5A and Supplementary Fig. S6A). We further sought to speculate on the total glycolytic activity in the two groups of cells (shFTH1 vs. SCR). A lower basal and compensatory glycolytic rate was detected in FTH1-silenced hESCs, suggesting that mitochondrial respiration is somehow inhibited in these cells which retain the capability to manage energy demand despite the genetic modi cation in the ferritin gene ( Fig. 5B and Supplementary Fig. S6B). For a more comprehensive metabolic investigation, we analyzed the expression levels of glycolysis-speci c genes such as ALDOA, PGK1, ENO1, PKM, together with genes encoding for enzymes of gluconeogenesis, G6PC1 and FBP1. In line with the results obtained by the glycolytic rate analysis, we could not observe signi cant changes in the expression of glycolytic genes between SCR and FTH1-silenced hESCs (Fig. 5C); on the other hand, the expression of G6PC1 (Glucose-6-Phosphatase Catalytic Subunit 1) resulted signi cantly higher in FTH1-silenced cells (Fig. 5D). G6PC1 encodes one of the key enzymes responsible for glucose production from glucose-6-phosphate, and therefore is critically involved in glucose homeostasis. This nding perfectly matches with the fact that high glucose levels are required for the proper functionality of detox systems, such as pentose phosphate pathway (PPP), glucuronidation pathway, and glutathione biosynthesis pathway, all of which are regulated by the Nrf2 signaling pathway (29,51). In support of this, we found that G6PD (glucose-6-phosphate dehydrogenase) and PGD (6-Phosphogluconate dehydrogenase), two key enzymes involved in PPP, were signi cantly increased in FTH1-silenced cells (Fig. 5E), strengthening the concept that the PPP may represent a crucial metabolic target by which the Nrf2 signaling potentiates the antioxidant response in hPSCs. The expression of PDK1 (pyruvate dehydrogenase kinase 1) was signi cantly enhanced in shFTH1-hESCs as well (Fig. 5F); its expression is positively regulated by HIF1 , previously shown to participate in ROS-induced metabolic reprogramming, where PDK1 prevents the pyruvate from entering the TCA cycle (tricarboxylic acid cycle or Krebs cycle) (40). We nally tested the expression level of UCP2 (Uncoupling protein-2), a protein of the inner mitochondrial membrane, involved in uncoupling OXPHOS from ATP synthesis. Similarly, to PDK1, UCP2 expression was higher in FTH1-silenced hESCs (Fig. 5G). As PDK1, UCP2 blocks pyruvate entry into the Krebs cycle and shunts it towards the PPP (31, 60, 85). Altogether, our ndings are indicative of a metabolic reorganization occurring in response to iron-mediated oxidative stress.

Discussion
Pluripotent stem cells (PSCs), including ESCs and iPSCs, possess the unique and extraordinary selfrenewal capability, while maintaining their pluripotency (58). In addition to these properties, PSCs also have the ability to give rise to almost any cell type, both in vivo and in vitro. On the bases of these key features, the PSCs technology holds enormous potential to attempt many research and biomedical applications, including drug screening, development, and repurposing (5,16,78,89), disease modelling (47,(54)(55)(56)70), cell therapy (82), aging (59), and regenerative medicine (57). Although signi cant progress has been made, many hurdles still need to be overcome to push PSCs technology towards the road to clinical practice. Apart from immunogenicity and genomic instability, undoubtedly representing the two major challenges, the successful differentiation and functional maturation of PSC derivatives must faithfully recapitulate all aspects of somatic cells and tissues, including the metabolic function. For instance, little is known about the role of ferritin heavy chain (FTH1), the major ferroxidase protein, in human ESCs homeostasis. Ferritin is composed of 24 heavy and light chains, whose core is able to sequester up to 4500 iron atoms (26). Ferritin heavy chain, in particular, is responsible for the oxidation of ferrous iron [Fe (II)] to less reactive ferric iron [Fe (III)] (81). Even though there is a wide literature suggesting that FTH1 is a major player in protecting cells against oxidative stress (9), so far, its role in embryonic stem cells has not been fully elucidated. As shortly stated above, ROS play a role as second messengers in embryo development (12,23), and their effects, bene cial or deleterious, depend on the embryo developmental stage. ESCs, compared to most terminally differentiated cells, are extremely resistant to hypoxic stress as they live and greatly proliferate at low oxygen concentration (14,64), suggesting that ESCs are equipped with a highly e cient antioxidant system (3,22). Because of its strong ferroxidase activity, FTH1 is supposed to be one of the major players in this system, maintaining iron in a safe and bioavailable form. To gain insights on the effects of FTH1-mediated control of ROSinduced cellular stress, here we stably silenced the FTH1 gene in H9 hES cell line using the shRNA strategy. Intracellular ROS production, measured in shFTH1-hESC by ow cytometry together with mitochondrial superoxide species, revealed that silenced cells have decreased levels of ROS accumulation. Similarly, mitochondrial membrane potential detected by TMRM assay, revealed that FTH1 silencing produces a decrease in mitochondrial membrane depolarization in shFTH1-hESCs compared to SCR control. By monitoring self-renewal and pluripotency, key ndings emerged: morphologically, shFTH1 hESCs were more attened and had less AP + cells compared to SCR control cells; the core pluripotency genes OCT4, NANOG, and SOX2 resulted signi cantly overexpressed in FTH1-silenced cells; Immunoblot analysis of Nanog protein, also revealed an increased expression in silenced cells. As pointed out elsewhere in the text, NANOG expression is under the control of the nuclear factor erythroid 2-related factor 2 (Nrf2), a stress-activated transcription factor involved in the cellular response to multiple injuries, among which oxidation is the most prominent. In a recent review, Dai and Coll. (10) summarize Nrf2-Keap1 signaling in the context of stem cell state and function, suggesting that it can serve as a master regulator of stem cell redox and metabolic homeostasis. Nrf2, in fact, can e ciently modulate ESCs and iPSCs pluripotency status by nely tuning NANOG and OCT4 expression through direct binding to upstream regions of these genes and delaying their proteasome-mediated ubiquitination. Our data showed clear evidence of the increased Nrf2 expression in hESCs, both at mRNA and protein level, upon FTH1 silencing. Overexpression of Nrf2 is therefore responsible for the upregulation of NANOG, as well as for the activation of the Gpx4 (glutathione peroxidase 4), ROS scavenger protein. A correlation between activation of Nrf2 and induced expression of FTH1 was previously reported in mouse embryonic broblasts (MEFs) (61). Ferritin is also known to function as an anti-apoptotic protein (4). Accordingly, our results show that FTH1-silencing triggers the activation of apoptosis-associated genes like BAX, BIM, CASP3 and CASP9; we further deepened our analysis by examination of the AKT and ERK1/2 pathways, which undergo signi cant perturbations during apoptosis. Previous studies have reported con icting results, showing that MAPK/ERK and PI3/AKT signaling pathways can be either activated (6) or decreased (2) in mESCs and hESCs upon ROS exposure. Our data indicate that FTH1 silencing produces a signi cant reduction of the pAkt, while activating ERK1/2 signaling. Concerning the metabolic point of view, most of the studies have focused on glycolytic metabolism and on its role in maintaining pluripotency as well as the importance of a metabolic shift of somatic cells to enter pluripotency during reprogramming; however, our study went beyond previous reports, speculating whether modulation of FTH1 by repression impacts hESCs homeostasis and metabolic properties. Interestingly, extensive analyses carried out showed that FTH1-silenced hESCs undergo metabolic reorganization. We observed that sh-treated hESCs cells respond to FTH1 silencing by activation of antioxidant systems including enzymes such as SODs, CAT, and GPXs. Moreover, FTH1-silenced hESCs had an increased expression of G6PD gene which is one of the major sources of NADPH, essential for maintaining the redox homeostasis (72). Together with G6PD, PDK1 and UCP2 were upregulated as well in silenced cells, suggesting a rerouting of metabolites from glycolysis to PPP and corroborated by a reduced glycolytic ux in shFTH1-hESCs. Consistent with the idea that PPP ux leads to an elevated reduction of NADP + to NADPH (21, 63), we suggest that an increased PPP ux is favored in hESCs with repressed FTH1 expression to drive the ROS clearance.

Conclusion
To our knowledge, this is the rst study uncovering the role of ferritin modulation by repression in human ESCs. In agreement with our work, previous studies have reported that ESCs are very resistant to oxidative stress. Interestingly, here we provide evidence of a potential interplay between the Nrf2 axis and oxidative pentose phosphate pathway (PPP) as molecular mechanism underlying the antioxidant response in hESCs in a shFTH1-induced stress condition (Fig. 6). Future investigations are necessary to validate the conclusion that can be drawn from this study. This manuscript does not report on the use of animal data.

Consent for Publication
Not relevant

Competing interests
The authors con rm that there are no con icts of interest.

Acknowledgments
Not applicable.

Data Availability
The processed and normalized data supporting the conclusions of this article are included within the article (Supplementary les). Raw data used during the current study are available from the corresponding author upon reasonable request.

Funding
This study was supported in part by grant # PRIN 2017CH4RNP_001 to G.C.
Authors' contributions GC and EIP conceptualized, designed and supervised the study; EIP, LS, and VL conceived the experiments; LS, VL carried out most of the experiments with the support of MLC and CZ; EIP, LS, VL, and SS analyzed the data; FB and AS performed cyto uorimetric analysis and analyzed the relative data; EIP, GC, and VL were involved in drafting the manuscript and reviewed its intellectual content; FSC critically reviewed the manuscript; GC was responsible for funding acquisition. All co-authors have reviewed the manuscript and approved its content prior to submission. All co-authors con rm that the manuscript has been submitted solely to this journal and is not published, in press, or submitted elsewhere.  and are represented as mean ± SEM (n = 3); signi cance is calculated by t-test: *P ≤ 0.05; **P ≤ 0.01, ***P ≤ 0.001. upregulation of both proteins in FTH1-silenced cells with respect to SCR cells. In all western blots presented in this gure, actin levels were evaluated to con rm equal loading control. For qRT-PCR data were normalized to value in SCR cells and are represented as mean ± SEM (n = 3); signi cance is calculated by t-test: *P ≤ 0.05; **P ≤ 0.01, ***P ≤ 0.001. Figure 4 genes in SCR and shFTH1-hESCs measured by qRT-PCR analysis. (E) Pentose phosphate pathway (PPP) genes G6PD and PGD were signi cantly up-regulated shFTH1 cells with respect to SCR control cells. (F) The PDK1 gene, responsible for inactivation of pyruvate dehydrogenase and (G) the gene UCP2, that uncouples OXPHOS from ATP synthesis, resulted both signi cantly up-regulated in FTH1-KD cells compared to SCR cells. For all qRT-PCR graphs, results were normalized to measurements in SCR cells (n = 3); *P ≤ 0.05; **P ≤ 0.01, ***P ≤ 0.001, t-test compared to SCR.

Figure 6
Schematic diagram illustrating oxidative stress response mediated by FTH1 downregulation in hESCs. FTH1 (1) is the master scavenger molecule for intracellular iron homeostasis and its inactivation leads to imbalanced redox homeostasis due to intracellular iron accumulation (2) and ROS production (3).
Surprisingly, in hESCs, a reduced expression of FTH1 expression is not associated with oxidative stress and ROS production. This phenomenon can be explained by three crucial ndings: rst, FTH1-silenced hESCs showed an enhanced transcriptional activity of Nrf2 (4-5) on key antioxidant target genes such as HIF-1, HMOX-1, NQO-1, GPX2/3/4, SOD1/2, FTL (6); second, we observed the activation of an alternative metabolic program from glycolysis (7-8) to pentose phosphate pathway (PPP) where the activity of the G6PD enzyme ensures the conversion of NADP+ to NADPH (9) involved in ROS neutralization; third, the mitochondrial metabolism was found to be inactive (10) in FTH1-silenced hESCs as demonstrated by the upregulation of PDK1 and UCP2 genes (11), speci cally involved in preventing pyruvate to enter into the tricarboxylic acid cycle (TCA) (12